effect of HT LINK SPEED on performance - myth or reality???

Urizen · Jan 30, 2008

here's the problem: I have an AM2 CPU (the 4400+) that's meant for HT2000 (that's an HT link speed of 1000mhz ie. HT multiplier of 5x)

but my mainboard only supports HT1600 maximum (ie. HT link speed of 800mhz maximum ie. multiplier goes up to 4x only)
yeah that means a crappy board but one of the only few that still supports AGP

which means my current HT link speed is 800mhz only
(my CPU still runs at its stock 2300mhz speed though)

now a few days ago I read this article which basically states that HT link speed has virtually no effect on performance (even 3D)
so I felt kinda relieved, sure I couldn't get my Ht link speed to 1000Mhz but apparently that wouldn't have changed anything performance-wise

but then I came across a topic on this forum, and specifically this post where the author clearly states that not only is there a difference in performance between 800 and 1000mhz, but also that the difference is HUGE, which completely contradicts the previous article

so the question (for expert experts only) is : which of the two is right ? in the article I linked the reviewer states (and explains) why HT link speed has little to no effect, yet here we see just the opposite !!! so is this paranormal or is there another explanation ? :banghead:

and if accordpower99's right and an HT of 800 really is crap, then this means my CPU is operating BELOW its potential and I really absolutely MUST find a way to bypass bios limitations & increase my HT speed to 1000mhz !
is there a windows utility that can do it ?
any help greatly appreciated !

hat · Jan 30, 2008

Can you drop your multiplier and raise the FSB to 250?

Darknova · Jan 30, 2008

Only in synthetic benchmarks does the HT link matter...even then it's no where near as much as you'd think he made it sound.

The HT link can go down as low as 750Mhz with a dual-core Athlon with no adverse effects.

You want to stay between 800 and 1000Mhz just to be safe though.

In the real world you will see no noticeable difference between 800Mhz and 1000Mhz

Hope that helps

AphexDreamer · Jan 30, 2008

When the guys says,

"no i finally found it man, my ht link speed at 1000mhz finally fixed it. I had it at 800 apparently, but eveythings great now!"

How do you know he is talking about performance increase, typical increasing the and bringing to normal the HT Link speed brings a huge amount of stability.

Maybe the guy was trying to overclock, it was not working, then brought the HT linkl speed to normal and it was stable again.

Simply said, HT link effect mostly Stabilty and practicly has no perfromance increase unless in help stabalize your CPU overclock in which case it aided to help increase the speed of the CPU whcih effected performance.

But, hey I'm no expert, thats just what I think is the case, besides I didn't even read the whole thread just one post that you gave me from it.

Mussels · Jan 30, 2008

back when i was AMD, i had my HT link from 2000 to 3000 (200 FSB to 300, without lowering it)

Wasnt very stable, but it wasnt any faster either - so i lowered it.

I dont think the speed difference will be that great, the guy who claimed it made a huge difference was probably from other factors - overclocking other components for example.

theonetruewill · Jan 30, 2008

I keep mine at 600, instead of 1000, because I actually find it faster- yeah it's odd but I have the proof with superpi. There is a pitiful synthetic performance difference for the people that have it higher (pm pos_pc) but there's no noticeable performance differnce in the realworld.

cdawall · Jan 30, 2008

i ran it between 600~1200 and i think i got a difference of .2s in superpi between the lowest and highest. it make almost no difference so dont worry about it.

trog100 · Jan 30, 2008

yes anywhere between 600 and 1200 wont show any real difference.. higher just brings in possible instability with no gain..

trog

Urizen · Jan 30, 2008

thanx for the positive replies folx

hat said:
Can you drop your multiplier and raise the FSB to 250?

maybe but I'm not sure what you mean by FSB here (looks like things are more complicated with recent AMD setups)

apparently there's two completely separate "base frequencies", one for RAM & one for the bus (HT)
then there's the CPU base clock but I'm not sure if it's the same as the RAM or the same as the HT base clock :confused:

here this is what Everest shows about my mainboard :

both ram and HT have a base clock of ~200mhz but it's not the same 200mhz :confused:

(the HT link is at 848mhz instead of 800 because I activated something called "AM2boost" option in bios)

Urizen · Jan 30, 2008

btw this is the topic I was talking about, it's still in the same page as mine, same section :
http://forums.techpowerup.com/showthread.php?t=50970&page=2

it's definitely a performance issue, at least that's what he says

Darknova said:
Only in synthetic benchmarks does the HT link matter...even then it's no where near as much as you'd think he made it sound.

well he did say it was "day & night" in terms of performance (he didn't know why his games ran a lot slower with his new 6400+, and he was even about to reformat his hdd, it was that bad) then he went from 800 to 1000 & according to him there was a huge boost

AphexDreamer said:
How do you know he is talking about performance increase, typical increasing the and bringing to normal the HT Link speed brings a huge amount of stability.
Maybe the guy was trying to overclock, it was not working, then brought the HT linkl speed to normal and it was stable again.

no he said it fixed the "slow" problem - besides if it were about stability only then 800 would've been better than 1000, no? instead he went up from 800 to 1000 :confused:

Mussels said:
I dont think the speed difference will be that great, the guy who claimed it made a huge difference was probably from other factors - overclocking other components for example.

that's what I thought (hoped) at first but in the topic - I linked to it again - he definitely says his games were running a lot slower depsite his new CPU before he made that simple change in the bios

so...how come?

accordpower99 · Jan 30, 2008

how about this tell me how to screen shot my bios, my mulitplier is at 16x bus speed 200, eveyrthing else auto except ht link.

nothing else has worked for me in overclocking thats why its still at stock settings. letme know how to screen shot if you dont believe me. :toast:

the night and day was the both games runing extemely choppy vs. nice and smooth after ht link set at 1000mhz

xvi · Jan 31, 2008

Download and run PerfMonitor from www.cpuid.com. Left click one of the bars and choose "HyperTransport bus 0 Bandwidth". If that maxes out, do anything you can to raise your FSB. Don't know what to say other than that.

Mussels · Jan 31, 2008

xvi said:
Download and run PerfMonitor from www.cpuid.com. Left click one of the bars and choose "HyperTransport bus 0 Bandwidth". If that maxes out, do anything you can to raise your FSB. Don't know what to say other than that.

you can SEE that? sweet! i'll try this out on my AM2 system.

with my housemate watching pokemon on the system, the HT0 went between 250 and 800 (of 2000)
Will have to test it with something else later, but that seems a great diagnostic and perfect for this thread. Thanks for linking that.

P.S the HT0 option is low down on the list - there are lots of options, so scroll down to find it.

trog100 · Jan 31, 2008

that amd boost thing is some kind of auto overclocking.. it caused me problems till i turned it off..

u have a baseline clock (fsb) 200mhz.. memory.. cpu speed and htt speeds are controlled by that..

htt is 5 x baseline clock = 1000..

memory would be 2 x baseline clock = 400/800

cpu would be 15 x baseline clock = 3000...

if everything is on auto raising the baseline clock will increase the speed of the other three things..

on manual u can independently adjust the speed of the other three things..

messing with the htt speed on its own will not make any real difference to performance.. but if u raise the baseline clock too much without lowering the htt multiplier to x 4 it might cause instability issues..

trog

Urizen · Jan 31, 2008

trog100 said:
messing with the htt speed on its own will not make any real difference to performance..

but then how d'you explain this :

accordpower99 said:
how about this tell me how to screen shot my bios, my mulitplier is at 16x bus speed 200, eveyrthing else auto except ht link.

nothing else has worked for me in overclocking thats why its still at stock settings. letme know how to screen shot if you dont believe me.

the night and day was the both games runing extemely choppy vs. nice and smooth after ht link set at 1000mhz

see folks that's what I meant - that's also what he said in the other topic, he only changed the HT speed (from 800 to 1000), nothing else

which basically contradicts everything that's said on the net about HT speed not affecting performance, because according to his results not only does HT link speed affect performance but it also has a HUGE effect

at least that's the only rational explanation, afaik :ohwell:

Mussels said:
with my housemate watching pokemon on the system, the HT0 went between 250 and 800 (of 2000)

yeah me too

but isn't that bad news ? cos' this also suggest that HT link affects performance (unlike what many ppl think)

Urizen · Jan 31, 2008

xvi said:
Download and run PerfMonitor from www.cpuid.com. Left click one of the bars and choose "HyperTransport bus 0 Bandwidth". If that maxes out, do anything you can to raise your FSB. Don't know what to say other than that.

yeah it maxes out but only during CPU-intensive stuff like winrar archiving (with multithreading option on)

maybe an HT link of 1000Mhz is necessary for dual-core systems ?

I tried raising the FSB in bios

now the thing is my CPU is meant for 200 x 11.5

so I raised the FSB to 220 and lowered the multiplier to 10.5

it booted and sure enough the HT link speed was 880mhz

but then anything above 220mhz won't boot ! :mad:

I tried 225x10 and it didn't boot
fortunately I had "boot failure guard" enabled, and when it did post then it showed a multiplier of...11.5 ! (and CPU frequency of 2.6Ghz or something !)
so basically when I raise the FSB above 220 then it ignores the manual multiplier settings & reverts back to the default 11.5 :confused:

so looks like I'm stuck with a maximum FSB of 220mhz thus maximum HT of 880mhz :banghead:

so is there any other way I can force an HT of 1000mhz ?

is there a utility to override bios HT multiplier settings & force it to 5x instead of 4x?

if I make a bios dump is there a simple way to add the 5x option ? (a bit crazy but I'm desperate now

)

trog100 · Jan 31, 2008

Urizen said:
but then how d'you explain this :see folks that's what I meant - that's also what he said in the other topic, he only changed the HT speed (from 800 to 1000), nothing else

which basically contradicts everything that's said on the net about HT speed not affecting performance, because according to his results not only does HT link speed affect performance but it also has a HUGE effect at least that's the only rational explanation, afaik

i would make no attempt to explain that.. i know what i know.. i dont give too much credence to one guys post as to what happens for him.. attempting to explain the inexplicable is a fools errand..

trog

ps..u could turn of the stupid auto overclock thing if u still have it running..

Urizen · Jan 31, 2008

trog100 said:
i would make no attempt to explain that.. i know what i know.. i dont give too much credence to one guys post as to what happens for him.. attempting to explain the inexplicable is a fools errand..

trog

ps..u could turn of the stupid auto overclock thing if u still have it running..

he'd have no reason to lie :confused:

unless there's another explanation (I hope)

anyway, I usually leave AM2boost (auto-overclock) on as I've had no stability probs with it (even with memtest)
HOWEVER I had to turn AM2boost off to be able to manually overclock (FSB, etc.) so the results I mentioned before (about system going unstable when FSB went above 220mhz) were with AM2boost OFF

btw the maximum vcore bios lets me set is 1.3V
now I suppose the necessary voltage depends only on the CPU's baseline clock (and not the "overall clock" which is baseline x multiplier), right ? in other words for example 200x1 requires as much voltage as 200x20 ? :confused:

if so maybe that explains why it wouldn't boot above 220mhz, if you need more than 1.3V for high FSB speeds

Mussels · Jan 31, 2008

CPU voltage depends on the overall clock, not the HTT speed.

Motherboard (northbridge) voltage may need to increase as the HTT goes up. - also you may be overclocking your ram, so it will need more volts too.

Urizen · Feb 1, 2008

so a CPU setting of, say, 200x15 will need exactly the same voltage as 300x10 ?

iirc there's no "northbridge" or "chipset" voltage option in bios, only vcore, dram & AGP voltage :confused:

btw. when I changed the FSB, the RAM clock was unchanged (it stayed at 400mhz (or 800mhz, whatever))

maybe it's an HT thing, maybe with hypertransport the chipset & RAM clocks are independent :confused:

PVTCaboose1337 · Feb 1, 2008

Ima give you a story. I have my CPU. I have the HTlink at like 500mhz during OCIng. I do some benchies. I raise the HT to 765. I do benchies. Almost the same. It is about a 3% difference for me. It is not a myth but on the quad core phenoms expect a 20% difference.

Darknova · Feb 1, 2008

Urizen said:
so a CPU setting of, say, 200x15 will need exactly the same voltage as 300x10 ?

iirc there's no "northbridge" or "chipset" voltage option in bios, only vcore, dram & AGP voltage

btw. when I changed the FSB I remember that the RAM clock remained unchanged (it stayed at 400mhz (or 800mhz, whatever) regardless of FSB clock

maybe it's an HT thing, maybe with hypertransport the chipset & RAM clocks are independant

Yes, because the CPU is still running at exactly the same speed as it was before. 3Ghz.

A lot of boards (especially cheaper ones) don't give you that many options for overclocking. We're talking in general terms, but if there is no option then there's nothing you can do about it unfortunately.

I think you'll find that unless you have the RAM divider set to Auto then the RAM clock will have changed, but most RAM will overclock to some degree, and you didn't push it very far.

And no, everything is dependant on your bus speed. CPU, RAM, HT, everything.

Urizen · Feb 1, 2008

well I set the FSB values from 200mhz up to 220mhz and each time the RAM clock remained unchanged @ 400mhz (according to cpuz & everest) :confused:

however activating AM2boost does increase my ram clock by about 5% (as well as HT clock) but also disables manual overclocking

I haven't overclocked my RAM though (I set my DRAM voltage to "auto")

however my RAM is certified for 5-5-5 and I set its timings to 4-4-4 & it still works without a problem

I dunno if decreasing timings counts as "overclocking" though

PVTCaboose1337 said:
Ima give you a story. I have my CPU. I have the HTlink at like 500mhz during OCIng. I do some benchies. I raise the HT to 765. I do benchies. Almost the same. It is about a 3% difference for me. It is not a myth but on the quad core phenoms expect a 20% difference.

thanks m8 that's good news ! :cool:

btw I just tried my own benches using 3Dmark05

I did all tests except the CPU ones (3Dmark05 doesn't use both cores)

I did the benchmarks using various HT link speeds from 200mhz up to 800mhz (checked in cpuz & everest to make sure new HT clock settings were applied)

the average FPS for most test didn't change at all regardless of HT clock

with the exception of one test, "vertex shaders - simple" (it's also 3dmark05's toughest test)
with this test, the average fps were as follows :
- HT clock @ 200mhz : 9fps
- HT clock @ 400mhz : 11fps
- HT clock @ 600mhz : 12fps
- HT clock @ 800mhz : 12fps (same as with 600mhz)

so apparently full potential is achieved at around 600mhz HT

I couldn't test it at 1000mhz but if performance is the same at 600mhz & 800mhz then it's likely it won't change at 1ghz

so you're probably right it only affects performance with quad-core & super-high-end CPUs but by the time I get one of these (not anytime soon) I'll have a new mobo & ram by then :cool:

PS. and I dunno what happened with accordpower99's setup but maybe it's cause he's got a top-line CPU over 3ghz and at those speeds HT link does have greater effect :confused:

accordpower99 · Feb 2, 2008

Urizen said:
PS. and I dunno what happened with accordpower99's setup but maybe it's cause he's got a top-line CPU over 3ghz and at those speeds HT link does have greater effect

OK UPDATE ON THIS I DUNNO WHAT MADE THE BIG CHANGE, I CHANGED IT BACK TO 800 AND GOT THE SAME PERFORMANCE, MAYBE THE AUTO WHICH KEPT IT AT 200? BUT THAT DOESNT MAKE SENCE, I DIDNT OC ANYTHING ELSE IM ALMOST POSITVE. SO TO EVERYONE :respect:

I SERIOUSLY THOUGHT THATS WHAT IT WAS, TRUST ME THERE WAS A BIG DIFFERENCE I JUST DONT KNOW WHAT IT WAS NOW. :confused:

accordpower99 · Feb 2, 2008

Another Question Will The Htlink Make The Cpu Run Hotter?

System Name	Starlifter :: Dragonfly
Processor	i7 2600k 4.4GHz :: i5 10400
Motherboard	ASUS P8P67 Pro :: ASUS Prime H570-Plus
Cooling	Cryorig M9 :: Stock
Memory	4x4GB DDR3 2133 :: 2x8GB DDR4 2400
Video Card(s)	PNY GTX1070 :: Integrated UHD 630
Storage	Crucial MX500 1TB, 2x1TB Seagate RAID 0 :: Mushkin Enhanced 60GB SSD, 3x4TB Seagate HDD RAID5
Display(s)	Onn 165hz 1080p :: Acer 1080p
Case	Antec SOHO 1030B :: Old White Full Tower
Audio Device(s)	Creative X-Fi Titanium Fatal1ty Pro - Bose Companion 2 Series III :: None
Power Supply	FSP Hydro GE 550w :: EVGA Supernova 550
Software	Windows 10 Pro - Plex Server on Dragonfly
Benchmark Scores	>9000

Processor	AMD FX 8320 @ 4GHz
Motherboard	Gigabyte GA-990FXA-UD5 rev1
Cooling	Corsair H70
Memory	4 x 4GB DDR3 Ripjawz 1600Mhz
Video Card(s)	Sapphire Vapor-X AMD R9 280X
Storage	1 x 500GB Samsung Evo 850, 1 x 500GB Vrap Data Drive, 3 x 2TB Seagate, 1 x 1TB Samsung F1
Display(s)	3 x DGM IPS-2402WDH
Case	Coolermaster HAF X
Audio Device(s)	Onboard
Power Supply	Coolermaster 1000W Silent Pro M
Mouse	Logitech G502
Keyboard	Logitech G510
Software	Windows 10 Pro x64

System Name	Unknown
Processor	AMD Bulldozer FX8320 @ 4.4Ghz
Motherboard	Asus Crosshair V
Cooling	XSPC Raystorm 750 EX240 for CPU
Memory	8 GB CORSAIR Vengeance Red DDR3 RAM 1922mhz (10-11-9-27)
Video Card(s)	XFX R9 290
Storage	Samsung SSD 254GB and Western Digital Caviar Black 1TB 64MB Cache SATA 6.0Gb/s
Display(s)	AOC 23" @ 1920x1080 + Asus 27" 1440p
Case	HAF X
Audio Device(s)	X Fi Titanium 5.1 Surround Sound
Power Supply	750 Watt PP&C Silencer Black
Software	Windows 8.1 Pro 64-bit

System Name	Rainbow Sparkles (Power efficient, <350W gaming load)
Processor	Ryzen R7 5800x3D (Undervolted, 4.45GHz all core)
Motherboard	Asus x570-F (BIOS Modded)
Cooling	Alphacool Apex UV - Alphacool Eisblock XPX Aurora + EK Quantum ARGB 3090 w/ active backplate
Memory	2x32GB DDR4 3600 Corsair Vengeance RGB @3866 C18-22-22-22-42 TRFC704 (1.4V Hynix MJR - SoC 1.15V)
Video Card(s)	Galax RTX 3090 SG 24GB: Underclocked to 1700Mhz 0.750v (375W down to 250W))
Storage	2TB WD SN850 NVME + 1TB Sasmsung 970 Pro NVME + 1TB Intel 6000P NVME USB 3.2
Display(s)	Phillips 32 32M1N5800A (4k144), LG 32" (4K60) \| Gigabyte G32QC (2k165) \| Phillips 328m6fjrmb (2K144)
Case	Fractal Design R6
Audio Device(s)	Logitech G560 \| Corsair Void pro RGB \|Blue Yeti mic
Power Supply	Fractal Ion+ 2 860W (Platinum) (This thing is God-tier. Silent and TINY)
Mouse	Logitech G Pro wireless + Steelseries Prisma XL
Keyboard	Razer Huntsman TE ( Sexy white keycaps)
VR HMD	Oculus Rift S + Quest 2
Software	Windows 11 pro x64 (Yes, it's genuinely a good OS) OpenRGB - ditch the branded bloatware!
Benchmark Scores	Nyooom.

System Name	COLOSSUS-MK4
Processor	E8400 @4.4 GHz - FSB @550 MHZ
Motherboard	Asus P5K Premium (Black Pearl)
Cooling	Xigmatek HDT-S1283
Memory	2x1GB Geil BlckDrgn 800 @1158 5-5-5-18
Video Card(s)	8800GT 512MB @740/1782/2080
Storage	Hitachi T7K250 250GB & 7200.10 Seagate 250GB
Display(s)	Gateway FPD1975W 19" Widescreen
Case	Antec 1200
Audio Device(s)	Xi-FI Xtreme Audio
Power Supply	CoolerMaster IGreen 500W
Software	XP Home SP3
Benchmark Scores	SuperPi: 10.563 Sciencemark: 2563.14

System Name	All the cores
Processor	2990WX
Motherboard	Asrock X399M
Cooling	CPU-XSPC RayStorm Neo, 2x240mm+360mm, D5PWM+140mL, GPU-2x360mm, 2xbyski, D4+D5+100mL
Memory	4x16GB G.Skill 3600
Video Card(s)	(2) EVGA SC BLACK 1080Ti's
Storage	2x Samsung SM951 512GB, Samsung PM961 512GB
Display(s)	Dell UP2414Q 3840X2160@60hz
Case	Caselabs Mercury S5+pedestal
Audio Device(s)	Fischer HA-02->Fischer FA-002W High edition/FA-003/Jubilate/FA-011 depending on my mood
Power Supply	Seasonic Prime 1200w
Mouse	Thermaltake Theron, Steam controller
Keyboard	Keychron K8
Software	W10P

System Name	money pit..
Processor	Intel 9900K 4.8 at 1.152 core voltage minus 0.120 offset
Motherboard	Asus rog Strix Z370-F Gaming
Cooling	Dark Rock TF air cooler.. Stock vga air coolers with case side fans to help cooling..
Memory	32 gb corsair vengeance 3200
Video Card(s)	Palit Gaming Pro OC 2080TI
Storage	150 nvme boot drive partition.. 1T Sandisk sata.. 1T Transend sata.. 1T 970 evo nvme m 2..
Display(s)	27" Asus PG279Q ROG Swift 165Hrz Nvidia G-Sync, IPS.. 2560x1440..
Case	Gigabyte mid-tower.. cheap and nothing special..
Audio Device(s)	onboard sounds with stereo amp..
Power Supply	EVGA 850 watt..
Mouse	Logitech G700s
Keyboard	Logitech K270
Software	Win 10 pro..
Benchmark Scores	Firestike 29500.. timepsy 14000..

System Name	CuStOm Built
Processor	AMD Athlon 64 X2 6400+ 16x, 211 bus, 3379mhz
Motherboard	MSI K9A Platinum
Cooling	COOLER MASTER Hyper TX2 CPU Cooler
Memory	CORSAIR XMS2 2GB 6400
Video Card(s)	ATI RADEON 256MB PCI-E X1950PRO 621/756mhz
Storage	COMING SOON WD 250GB SATAII 16mb cache
Display(s)	22"
Case	Custom
Audio Device(s)	creative xgamer live
Power Supply	ultra xfinity 500w
Software	win xp
Benchmark Scores	3dmark06 results 5245

System Name	Rainbow
Processor	Intel Core i7 8700k
Motherboard	MSI MPG Z390M GAMING EDGE AC
Cooling	Corsair H115i, 2x Noctua NF-A14 industrialPPC-3000 PWM
Memory	G. Skill TridentZ RGB 4x8GB (F4-3600C16Q-32GTZR)
Video Card(s)	ZOTAC GeForce RTX 3090 Trinity
Storage	2x Samsung 950 Pro 256GB \| 2xHGST Deskstar 4TB 7.2K
Display(s)	Samsung C27HG70
Case	Xigmatek Aquila
Power Supply	Seasonic 760W SS-760XP
Mouse	Razer Deathadder 2013
Keyboard	Corsair Vengeance K95
Software	Windows 10 Pro
Benchmark Scores	4 trillion points in GmailMark, over 144 FPS 2K Facebook Scrolling (Extreme Quality preset)

System Name	Whim
Processor	Intel Core i5 2500k @ 4.4ghz
Motherboard	Asus P8Z77-V LX
Cooling	Cooler Master Hyper 212+
Memory	2 x 4GB G.Skill Ripjaws @ 1600mhz
Video Card(s)	Gigabyte GTX 670 2gb
Storage	Samsung 840 Pro 256gb, WD 2TB Black
Display(s)	Shimian QH270 (1440p), Asus VE228 (1080p)
Case	Cooler Master 430 Elite
Audio Device(s)	Onboard > PA2V2 Amp > Senn 595's
Power Supply	Corsair 750w
Software	Windows 8.1 (Tweaked)

effect of HT LINK SPEED on performance - myth or reality???

New Member

Enthusiast

Freshwater Moderator

where the hell are my stars

New Member

New Member

New Member

Freshwater Moderator

New Member

New Member

New Member

Freshwater Moderator

New Member

Graphical Hacker

New Member

New Member

New Member